I want to launch this topic by asking a music library-related question pertaining to Classical Music, but it needn’t be an exclusively classical issue.
A lot of classical albums feature lengthy pieces that are broken down on CDs - often quite arbitrarily - into multiple tracks, each of which are merely fragments of the piece, and are typically not self-contained works that you would want to play in isolation. A great example is Mahler’s 8th symphony, which comprises two lengthy movements, each usually broken down into a dozen or more separate tracks. This can have the effect of making the user’s music library more cumbersome than it needs to be. Opera is a genre where this issue is particularly common.
One solution is to rip those individual long movements into one large single track, or alternatively, combine the tracks into one after ripping or downloading. I do that widely throughout my own music library, but I would really like to know how many of you do this. Am I an outlying case, or is this a widespread practice? It may make a difference to Octave if this turns out to be a widely-used practice.
Using the Mahler as an example, it is indeed in two parts but the separate tracks within these two parts are not arbitrarily placed but are specific individual portions of the work. Bach’s cantatas are like this, etc. These separate sections are all identified in the score with specific titles, performance directions, and more.
I want all of these tracks kept separate and individually identified when ripping and when placed in a music library. Not only are they individual sections of the work that I know, I do access them directly at times - such as when comparing interpretations of various performers, when I just want to hear a specific aria, when I am preparing to perform the work, etc.
In a pop context, Dark Side of the Moon is intended to be listened to as a whole and could also be concatenated into one single track, but I assume listeners want access to each individual track as well.
I rarely even look for index/sub-track marks in CDs, although I have some where they are used as markers for points the commentator is making in his discussion of the work in the booklet which came with the CD. Read the text, listen to the music at the index point, read some more, etc. It is great for studying a fugue and other complex music.
I have never checked if ripping software recognizes these marks.
Using the Mahler as an example, it is indeed in two parts but the separate tracks within these two parts are not arbitrarily placed but are specific individual portions of the work. Bach’s cantatas are like this, etc. These separate sections are all identified in the score with specific titles, performance directions, and more.
I want all of these tracks kept separate and individually identified when ripping and when placed in a music library. Not only are they individual sections of the work that I know, I do access them directly at times – such as when comparing interpretations of various performers, when I just want to hear a specific aria, when I am preparing to perform the work, etc.
What I find in my Mahler 8th collection is that the 1st movement is typically divided into either 6 or 8 tracks. While it's true that they use division points marked in the score, I often find that the individual tracks are found to have been given all sorts of different titles, so it can be unclear at a glance whether or not they represent identical (or at least equivalent) passages. I suppose one day I should set aside the time to take a listen and find out. As to the 2nd movement, that gets divided in far more varied manner from recording to recording, with anywhere from 12 to 18 separate tracks.
I’m not sure how this plays out in regard to Bach’s cantatas, but as regards Opera, it is generally much worse.
In the context of Octave, identifying and properly representing recordings of Works like Mahler’s 8th will depend in part on knowing, for example, whether we are likely to encounter a lot of users who have concatenated multiple tracks together … for whatever reason.
Just for clarity, Octave will never attempt to combine multiple tracks into one. The discussion relates only to the prospect of Octave being asked to import a User’s existing library where this may have been previously done by the User, in which case Octave will need to be able to figure out what it is that it is importing.
If a user combined tracks after ripping (as you describe in your first post) what alternate approaches could Octave take?
If Octave is confronted with a concatenated track, Octave will see a single labeled track (e.g., Mahler 8, Part II) the duration of which is much longer than any track on the corresponding CD. The metadata will often similarly not match. I would expect Octave to leave the user’s labeling alone and to import the track just as it is. That is, if something does not make sense the default would be to shrug, nod, and accept.
The issues which concern me here arise when Octave sets about identifying the specific recording, so that it can reach out and obtain all sorts of additional information which may not be present in the User’s metadata. There are tools Octave uses to do that, but those tools can fail from time to time - one of those times is when track concatenation has taken place. Identifying the Work itself from the metadata is not as straightforward as you might think. Sure, Mahler only wrote one 8th Symphony so that ought not to be an insurmountable challenge, but take for example Stravinsky’s Firebird. There is the original Firebird ballet, plus various versions of the Firebird Suite. If, for example, your metadata only identifies it as “Firebird”, what is Octave to do? Is it the ballet, or one of the suites? Or is it any one of a number of unrelated songs called Firebird, by such luminaries as White Noise, Bob Dylan, Star Inc., Taku Iwasaki, Gareth Emery, Galantis, or Eyot. How literally do we take the User’s metadata … do we account for typos? For example, is it possible that the Work is actually “Firebirds” by Clutch? Then, of course, the Firebird is also known as l’Oiseau de Feu. And so on …
Somewhat surprisingly, since I am quite anal about keeping my own metadata in order, my Octave prototype has already turned up a couple of albums I have in which the tracks turn out to have been wrongly labelled, so that it has alerted me to problems in my own metadata.
The issues which concern me here arise when Octave sets about identifying the specific recording, so that it can reach out and obtain all sorts of additional information which may not be present in the User’s metadata.
This has got to be tricky, even without concatenated tracks. Even in pop music this is difficult - consider all of the versions of Kind of Blue, Waltz for Debby, Dark Side of the Moon. Users will want the various editions, remasters, etc. correctly identified.
Are the tricks Octave can employ if a user has concatenated tracks? This seems like quite a roadblock to me.
By the way, I do not mind entering a search for the correct album if the search function and database are both robust. I know what I have so matching things up is easy if the search results are displayed well so I can pick the correct one.
